Google DeepMind has introduced a new artificial intelligence model that can determine how changes in DNA affect different types of cells. It allows scientists to investigate the genetic causes of diseases in greater detail and identify new avenues for treatment.
AlphaGenome 2 is an updated version of DeepMind's previous AlphaGenome model, introduced last year. Almost all cells in the human body have the same DNA, but different genes are switched on or off in different cells, depending on their type and state. As a result, the same genetic change can have different effects on heart muscle, skin or immune cells.
The new model can predict how genetic changes affect specific cell types and states, revealing consequences that more general models may fail to detect. It was trained on more than 100 million individual cell profiles from 68 tissues. According to DeepMind, AlphaGenome 2 identifies nearly twice as many genetic variants that may alter gene activity in cells as the previous best-performing model.
The model can also make predictions for cell types and biological states on which it was not previously trained. It can also identify gaps in experimental data, helping scientists determine what needs to be investigated in the laboratory.
According to Adam Gaioso, a Google DeepMind researcher involved in the project, the model's main achievement is connecting human genetics with cell biology. Ziga Avsec, the company's head of genomics, noted that this level of precision is important for treating diseases because drugs typically target specific proteins and cell types.
The significance of this model lies not only in its ability to identify more genetic changes. It is equally important for scientists to understand what those changes do in specific cells, because the mechanisms behind a disease cannot always be explained by looking at the overall picture of DNA alone. This distinction could eventually help researchers develop more targeted treatments.
Related News
However, there is also a practical limitation. More accurate predictions require substantial computing resources, meaning the model cannot yet replace previous tools in all research projects. Its performance also depends on the data used to train it. If DeepMind succeeds in filling existing data gaps and expanding its research across different species, systems like AlphaGenome could play a greater role in biology. Until then, their results should be viewed as a research tool rather than a ready-made solution for diagnosing or treating diseases.
